The actions of melatonin are multiple and many must derive essentially from modification of events in the CNS. However numerous melatonin target sites also exist in the periphery. Any endogenous free-radical scavenging activity does not require a receptor. Lesions of the SCN and the anterior hypothalamic area can block photoperiodic and/or circadian effects of melatonin in some rodents, but with a degree of disparity between laboratories (149). Implants or infusion of melatonin in the hypothalamus mimic or block photoperiodic responses in several species (41). In prepubertal rats melatonin inhibits GnRH-induced LH release in pituitary cultures at concentrations comparable to those circulating in the blood (47), and there is evidence that melatonin influences GnRH secretion from the hypothalamus (150).

Using 2-125I iodomelatonin as a ligand, high-affinity (Kd 25 to 175 pM), saturable, specific, and reversible melatonin binding to cell membranes was initially reported in the SCN (151) and the pars tuberalis of the pituitary (152). Subsequently binding has been found in many brain and other areas including cells of the immune system, a number number of cancer cell lines, the gonads, the kidney and, importantly, the cardiovascular system. The SCN shows clear binding in human postmortem tissue (153). Species variation of melatonin-binding sites in the brain is of course apparent. The most consistent (but not universal) binding site between mammalian species is the pars tuberalis. There is good evidence that the pars tuberalis transduces the effects of photoperiod, via melatonin, on seasonal variations in prolactin secretion in ruminants (154). Morgan (155) proposed that pars tuberalis cells secrete an entirely new hormone 'tuberalin' that subsequently mediates the physiological effects of melatonin-although to date the structure has not been elucidated. Pars distalis binding is absent in adult rats but persists after birth in the neonate (156). This suggests that binding may indeed underlie function, as melatonin inhibits GnRH induced pituitary LH release in prepuberty but not in adulthood. Moreover, binding is detectable in the brain of neonatal Syrian hamsters whose circadian system responds to melatonin whereas it is lost in adults who do not respond. There are also changes with time of day, with season and as a function of exposure to melatonin (see (12, 180) for references).

White and co-workers initially demonstrated that melatonin-induced pigment aggregation in amphibian melanophores is a pertussis toxin-sensitive system and that melatonin inhibits forskolin-activated cAMP formation (157). Intensive investigation of the properties of the pars tuberalis binding site has revealed that physiological doses of melatonin inhibit forskolin-activated cAMP production in vitro in a time- and dose-related manner (158, 159). Dubocovich and co-workers have demonstrated a functional melatonin receptor, initially in rabbit and chicken retina (inhibition of calcium-dependent dopamine release), which is localized in the inner plexiform layer containing dopamine amacrine cells in rabbits, in the outer and inner segments in mice, and possibly in the pigmented layer in some mammals (160). Nuclear melatonin receptors (RZR/ROR alpha and RZR beta) have been described and may be involved in peripheral melatonin effects (161). Genetic polymorphism has been identified within melatonin membrane receptors and further investigation of these polymorphisms in relation to photoperioidism, human disease, sensitivity to melatonin etc. is ongoing (162, 163). Melatonin membrane receptors have now been cloned and three initial subtypes were named Mel 1a, Mel 1b and Mel 1c (164). The Mel 1a receptor gene has been mapped to human chromosome 4q35.1. Its primary expression is in the pars tuberalis and the SCN. Of particular interest is the observation that melatonin can alter the expression of clock genes within the pars tuberalis in a manner analogous to photoperiod (174). Mel 1b has been mapped to chromosome 11q21-22 and its expression is in the retina and the brain. Mel 1c is not found in mammals. Two cloned mammalian receptors (Mel 1a, Mel 1b) have recently been renamed MT1 and MT2 (160). They are a new family of G protein coupled receptors, have high affinity (Kd 20-160 picomolar) and inhibit forskolin-stimulated cyclic AMP formation. MT1 acts through both pertussis sensitive and insensitive G proteins. The tissue expression in the SCN and the PT suggests that the circadian and reproductive effects are mediated through this receptor. However, using gene knockout technology and pharmacological manipulations, the results to date suggest that the phase shifting receptor is MT2, whilst MT1 is associated with acute suppression of SCN electrical activity in addition to its actions within the pars tuberalis. Several other physiological responses have been ascribed to MT1 and MT2 receptors, including (MT1) melatonin-mediated potentiation of adreneregic vasoconstriction and (MT2) modulation of dopamine release in the retina. A third putative mammalian melatonin receptor (MT3) remains somewhat controversial. A recent review summarises melatonin receptor pharmacology (160).

Large numbers of putative and actual melatonin agonists together with some antagonists have now been described e.g. (165). Most data is available concerning a series of agonists developed from napthalene derivatives. They show a range of affinity for the pars tuberalis melatonin receptor, some being of much higher affinity than melatonin. The most interesting have similar effects to melatonin on rhythm physiology in both rodents and humans, and antidepressant properties are emerging (166-173)

It is likely that SCN receptors mediate the circadian effects of melatonin, those in the mediobasal hypothalamus and pars tuberalis influence photoperiodic seasonal reproduction with regard to gonadotrophin secretion and prolactin respectively, and those in the retina mediate the retinal processes influenced by melatonin. The physiological functions of the multiplicity of melatonin binding sites in other areas remain to be clarified.

Probably the most interesting development in the mechanistic aspects of the effects of melatonin concerns its influence on gene expression in the pars tuberalis. Many clock genes are expressed in the pars tuberalis (Bmal1,Clock, Per1 Per2, Cry1, Cry2) with a 24h rhythmicity different from their expression in the SCN. Per1 is activated at the beginning of the light phase and Cry1 at the beginning of the dark phase. Long or short photoperiod information is encoded within the SCN. Melatonin synthesis, driven by the SCN, conveys this photoperiodic information to the pars tuberalis by virtue of its pattern of secretion. This in turn influences the pattern of expression of the clock genes per1 and cry1 within the pars tuberalis providing a means of translating the melatonin signal for the control of seasonal prolactin variations (174-176). Melatonin target sites in the hypothalamus influencing seasonal variations in reproductive hormones have yet to be fully defined and so far melatonin does not appear to influence clock gene expression in the SCN (177, 178). However, it has been proposed that other ‘calender’ cells will be identified in the CNS which regulate seasonal changes other than prolactin and may use the relative phasing of clock gene expression for translating the photoperiodic (melatonin) signal (179).

In rodent pars tuberalis cells rhythmic expression of per1 appears to be dependent on sensitization of adenosine A2b receptors which in turn depend on melatonin activation of MT1 receptors (180). Clearly it is possible that the melatonin signal is a widespread humoral mechanism related to biological timing, acting through modification of clock gene expression . It appears not to be of major importance to rhythm generation in the SCN but it is within the peripheral pars tuberalis system. The effects of melatonin on peripheral, as well as central, clock gene expression is likely to be a rich field of enquiry.